Mobile Biometric Identification Using Segmented Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image recognition technologies, such as YOLO, require significant computer resources and are not suitable for mobile devices, especially for real-time identification of specific biometric features like faces or palms, due to their complexity and resource intensity.
Innovation Solution
A method using a neural network on a mobile device to process images from an optical sensor, identifying biometric characteristics like palmar creases, irises, or footprints, by separating images into grid cells and using depthwise convolutional layers and batch normalizers to reduce resource requirements, allowing for efficient identification and storage of biometric features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional YOLO neural network is used for object identification, then identification accuracy is maintained, but computer resource consumption increases significantly making it unsuitable for mobile devices
Solution Approach 1:
The patent divides the image processing task into two distinct stages: a coarse detection stage that quickly identifies potential regions containing biometric features, and a fine extraction stage that precisely extracts the biometric characteristics from those regions. This segmentation allows the system to use a simplified neural network for initial detection, reducing computational load while maintaining accuracy through subsequent precise extraction.
Solution Approach 2:
The patent extracts only the essential biometric features (such as facial landmarks, fingerprint patterns, or iris characteristics) from the identified regions rather than processing the entire image at high resolution. By taking out and focusing only on the relevant biometric portions, the system achieves accurate identification with significantly reduced computational resources compared to full-image processing.
2Productivity
If complex neural network is used for real-time object identification, then identification speed improves, but device complexity increases making it unsuitable for mobile devices
Solution Approach 1:
The patent segments the neural network into a lightweight detection model for rapid object localization and a separate feature extraction model for biometric analysis. This allows the mobile device to run the simpler detection model in real-time for fast identification, while the more complex feature extraction can be performed with lower computational demands on the already-identified regions.
Solution Approach 2:
The patent applies partial action by using a simplified neural network that processes only the most critical features necessary for rapid identification, rather than analyzing all possible image features. This partial processing approach achieves sufficient identification speed for mobile devices while keeping the neural network complexity manageable.
Data Source
AI summary
A method includes obtaining, by a processing device from an optical sensor of a mobile device, an image; processing, by the processing device, the image by using a neural network to identify a position of an object in the image and the object in the image, thereby obtaining an identified object; after processing the image, extracting, by the processing device from the identified object, a biometric characteristic, and providing, by the processing device, at least the biometric characteristic as input to determine whether the biometric characteristic identifies a user.